Assessing Output in CoCounsel and Protégé

Chart the path to confident output in your AI legal research journey. This is the do-it-yourself version of the live workshop: the slides, the speaker notes, and the two things worth taking away — the VET process, and a checklist for reading an AI answer critically.

Presented: Thursday, 12 February 2026, on Zoom
Presented by: Heather Joy (hnjoy@law.stanford.edu) and Sergio Stone (sstone2@law.stanford.edu), Robert Crown Law Library
Format: Slides with speaker notes, worked through at your own pace

This was accurate when presented. CoCounsel and Protégé both change quickly, so what you see in the tools today may differ from what the slides show.

What the session covers

1. Scope and limits

What AI output in Protégé and CoCounsel can and cannot be relied on to do.

2. Evaluate

Reading the answer, the precedent it rests on, and the sources it consulted.

3. Verify

Confirming the output with the legal research tools you already know how to use.

Start with VET

The three-step habit the session is built around:

V — verify that the sources exist.
E — evaluate the content of each source, checking for the nuance the AI may have flattened: dates, jurisdiction, red flags, a summary of law that has since been overruled.
T — test the reasoning. Does the argument hold together? What is it assuming, and are those assumptions sound?

The evaluation checklist

Nine things to look at in any AI legal research answer:

Watch the sequence of sources it consulted
Scrutinize the structure of the answer
Be skeptical of the output’s tone
Review the precedents
Interpret the citator signals
Check for a visualization of the case law
Validate that the facts in your prompt were actually addressed
Confirm the categories of secondary sources
Watch how it handles questions spanning federal and state law

The presentation

Open the slides and follow along. The explanations are in the speaker notes, and the audio and video in the deck illustrate the points as you go.

Download the deck (.pptx) Open in Google Slides (opens in a new tab)

For the speaker notes, use the three-dot menu in the player above and choose Open speaker notes.

Privacy and confidentiality. Check how any AI tool treats your data before you upload anything to it. The rule of thumb from the session: if you would not want it on the front page of the newspaper, do not upload it. The PAUSE Rule is the longer version of that test.

Questions about any of this? Bring them to the AI Curiosity Corner, or email library@law.stanford.edu. All past sessions are in the archive.